CT4032

Mathematical Optimisation
 •Mathematical optimization,
What is often simply referred to as
Mathematical optimization, is a branch of
mathematics and computer
Optimisation science that deals with finding
the best solution among a set
of feasible alternatives. The ^^^
goal is to identify the values of
variables that optimize
(maximize or minimize) a
Number
given objective function while
satisfying a set of constraints.
vvv
Optimisation
Problem

The optimization problem is termed as

mathematical programming, which is the
mechanism through which we can find an
element, variable or quantity to best fit a set
of given criterion or constraints.
key concepts associated with mathematical
optimization
• Objective Function: This is the function that needs to be optimized. It represents the quantity you want to
maximize or minimize.

• Decision Variables: These are the variables that can be adjusted or varied to optimize the objective
function.

• Constraints: These are conditions or limitations that must be satisfied while optimizing the objective
function. Constraints could be limitations on resources, budget constraints, or any other relevant
restrictions.

• Feasible Region: The set of all possible combinations of decision variable values that satisfy the constraints.

• Optimal Solution: The set of values for the decision variables that provides the best (maximum or
minimum) value for the objective function within the feasible region.
 Example-1
• Problem: A company manufactures two products, X and Y. The profit per unit for X is $5, and for Y
is $8. The company has constraints on the availability of resources: it can produce at most 100
units of X, 80 units of Y, and it has 200 hours of labor available. Formulate an optimization
problem to maximize the profit.
 Example2: Cost Minimization for a Fence
• A farmer wants to build a rectangular fence to • Explanation:
enclose a grazing area for cattle. The farmer has • The objective is to maximize the area (A) of the
100 meters of fencing material. Find the
dimensions of the rectangular fence that would
rectangular fence.
maximize the enclosed area. • The constraint ensures that the total amount of
• Variables:
fencing material used (the perimeter of the
rectangle) does not exceed 100 meters.
• Let x be the length of the rectangle and y be the • Now, the farmer wants to find the values of x
width.
and y that satisfy the constraint and maximize
• Objective Function: the area.
• Maximize the area of the rectangle: A=x⋅y
• This is a simplified optimization problem where
the farmer is trying to use a fixed amount of
fencing material to enclose the largest possible
grazing area. It's a common type of problem
found in introductory optimization examples.
 Applications
• Operations Research:
• Supply Chain Management: Optimizing the distribution of goods and
minimizing transportation costs.
• Production Scheduling: Determining the most efficient schedule for
manufacturing processes.
• Finance:
• Portfolio Optimization: Selecting the optimal mix of financial assets to
maximize returns or minimize risk.
• Option Pricing: Determining the best strategy for options trading.
 Applications
• Engineering:
• Structural Design: Optimizing the design of structures for maximum strength with
minimum material usage.
• Network Design: Finding the most efficient layout for communication or transportation
networks.
• Telecommunications:
• Network Optimization: Optimizing the layout of communication networks for efficient
data transfer.
• Healthcare:
• Treatment Planning: Optimizing treatment plans in medical fields, such as radiation
therapy.
• Resource Allocation: Efficiently allocating healthcare resources, like hospital beds and
staff.
 Applications
• Agriculture:
• Crop Planning: Optimizing the allocation of land and resources for different
crops.
• Harvest Scheduling: Determining the best time to harvest crops for maximum
yield.
• Machine Learning and Data Science:
• Hyperparameter Tuning: Optimizing the hyperparameters of machine learning
models for better performance.
• Feature Selection: Identifying the most relevant features in a dataset for
model optimization.
Trying to Optimise
There are three core components when
performing optimisation:

• Objective function: - f(x)

• Variable - x1, x2, x3, x4

There can be none, one, or multiple variables.

• Constraints
• Equality - hn(x)
• Inequality - gn(x)
Minimisation &
Maximisation

The process of getting the

functional objective (number) as
low or as high as possible, which
can be done through changes to
the many variables.
 • Global Optimum
• The global optimum is the best possible solution across the
entire feasible region.
• It represents the lowest (in the case of minimization
problems) or highest (in the case of maximization
problems) value of the objective function over the entire
Global & solution space.
• Finding the global optimum ensures that no other feasible
Local solution provides a better objective function value.
• Local Optimum:
Optimum
Points • A local optimum is the best possible solution within a
specific neighborhood of a point in the solution space.
• It represents a solution where the objective function is
optimal given the constraints, but it may not be the best
solution globally.
• There can be multiple local optima in a problem, each
corresponding to a different region in the solution space.
Global &
Local
Optimum
Points
 Static & Dynamic Problems
Static Problems:
Definition: In static problems, the system or phenomenon under
consideration does not change with respect to time.
Characteristics:
All relevant parameters and constraints are constant.
The solution is sought at a specific point in time.
Changes in the environment or system are not considered during the
problem-solving process.
Examples:
Solving a system of linear equations.
Optimization problems with fixed parameters.
Structural analysis of a building under constant loads.
 Static & Dynamic Problems
Dynamic Problems:
Definition: Dynamic problems involve elements that change or evolve
over time.
Characteristics:
Parameters, variables, or constraints may vary with time.
Solutions are sought across different points in time.
Consideration of how the system evolves or responds to changes is
essential.
Examples:
Economic models that account for changes in market conditions over
time.
Traffic flow optimization considering varying traffic patterns.
Climate models predicting weather conditions over time.
 Stochastic & Deterministic
• Stochastic and deterministic are terms used to describe two different types of
mathematical and statistical models, particularly in the context of optimization and
decision-making.

• Deterministic Models:
• Definition: Deterministic models are those where the outcome is completely determined by the
input parameters and initial conditions. There is no randomness or uncertainty involved in the
model.
• Characteristics:
• The model has fixed and known input parameters.
• The same set of inputs will always produce the same output.
• There is no consideration of variability or randomness.
• Examples:
• Solving systems of linear equations.
• Optimization problems with known and constant parameters.
• Classical physics equations describing motion under certain conditions.
 Stochastic & Deterministic
• Stochastic Models:

• Definition: Stochastic models involve randomness or uncertainty in the

system. The outcome is not entirely predictable, and the model incorporates
probability distributions for one or more parameters.
• Characteristics:
• Input parameters are uncertain and described by probability distributions.
• The model produces different outcomes for the same set of inputs due to randomness.
• Stochastic models are often used to capture variability in real-world systems.
• Examples:
• Monte Carlo simulations for risk analysis in finance.
• Weather forecasting models that incorporate probabilistic predictions.
• Stochastic optimization models considering uncertain parameters.
Linear &
Nonlinear

Linear equations are having a

single degree and demonstrated
by a straight line, and nonlinear
can have two or more degrees and
are curved.
Mathematical Optimisation
Methods

Non-calculus
• Linear Programming
• Sawtooth Method

Calculus
• Newton’s Method
• Penalty Functions
• Lagrange Multipliers
Computational Optimisation
Techniques and Algorithms

DANTZIG’S SIMPLEX COMBINATORIAL QUANTUM

ALGORITHM ALGORITHMS OPTIMIZATION
ALGORITHM
Buying a House
Example

Think about the variables and

constraints regarding this.
If you want code & work on
assignment, you can
Any Questions